Open uphlewis opened 1 year ago
I see the same issue
Same here
Same issue here, in version 4.22.0 (117440) of Docker Desktop.
@sonbui00 @sebschlo @jvipez
It might be helpful for you to also generate a diagnostics ID for the devs to see how our setups intersect ?
https://docs.docker.com/desktop/troubleshoot/overview/#diagnose
@uphlewis C66ED105-8E20-44CC-A06D-8DFD6256C35A/20230822041856
@uphlewis 0DFD0413-58B0-4097-A9C1-461BC37DDE0F/20230822164831
@uphlewis D9D1B853-7EF6-40A3-BF0C-42486C35AA8A/20230822230010
Same, 4.22.1 on mac
Same here, 4.22.1, macOs Ventura 13.6
Upgrade to 4.23 or 4.24.
Same here on 4.24.0 on MacOs Monterey (Diagnostics ID: 82CFEBE5-ED2F-4CA3-B3CF-E41678370A02/20231005105346
)
In my case I had already paused a Docker Compose stack. Attempting to resume it after noticing Docker had unpaused led to Error response from daemon: Cannot unpause container: OCI runtime resume failed: container not paused: unknown
and a dangling image.
@rhysb27 - trying to understand your use case for pause/unpause - why exactly are you pausing the engine? In the case of resources optimization does the Resource Saver (which automatically does the job) work for you? Or are there any other use cases?
Just had this again on 4.24.0 (122432)
I paused docker, plugged in my laptop, checked power consumption / charging time estimation and it still listed docker as consuming a lot of power so i checked again and it was unpaused!
Mine resumes almost immediately, in 5 secs or so. I tried many times, with UI open & not, same behavior.
On latest Docker Desktop 4.24.2 on latest Windows 10 (as of 2023-10-23)
Same here, 4.24.2 Mac Os 13.6 (22G120) Docker pauses and resumes in under 5 seconds.
Same here too, Docker v4.25.0 on W Version 10.0.22621 Build 22621.
Engine restarts itself 2~5 seconds after pause command is given.
@rhysb27 in your diag I saw that you are running docker events
command. That will wakeup Docker from pause state. Can you try after terminating docker events
command and see pause state is retained ?
If it was not you who started docker events command, can you kindly check the parent process of that command and let us know which process started it ?
Could it be that intelliJ or VSCode with Docker plugin will trigger the resume?
It looks like the Docker gets unpaused any time when the "docker" command is used, doesn't matter if it is CLI , k3d or IntelliJ Docker plugin is connected. For me, the Docker can be in a paused state for quite a long period of time, but if the Idea is connected to it, Docker resumes in seconds after the pause button is pressed
my laptop's battery dies quickly if docker is enabled unintentionally 😀, at least it should give desktop notification so that i can pause it or fetch power cord.
when something is trying to access docker, i want it to say docker is not running
rather than unpause automatically.
This happens to me as well, gets unpaused after about 5s. Enabling debug logging shows the following:
log enabled with these settings:
{
"debug":true,
"log-level": "debug"
}
[2024-02-23T19:08:08.052146130Z][dockerd][I] time="2024-02-23T19:08:08.050975755Z" level=debug msg="Calling HEAD /_ping"
[2024-02-23T19:08:08.052151588Z][dockerd][I] time="2024-02-23T19:08:08.051076255Z" level=debug msg="Calling HEAD /_ping"
[2024-02-23T19:08:08.057347755Z][dockerd][I] time="2024-02-23T19:08:08.057276421Z" level=debug msg="Calling GET /v1.44/version"
[2024-02-23T19:08:08.057513463Z][dockerd][I] time="2024-02-23T19:08:08.057417088Z" level=debug msg="Calling GET /v1.44/version"
[2024-02-23T19:08:08.061171505Z][dockerd][I] time="2024-02-23T19:08:08.061109713Z" level=debug msg="Calling GET /v1.44/events?filters=%7B%22type%22%3A%7B%22container%22%3Atrue%7D%7D"
[2024-02-23T19:08:08.061200421Z][dockerd][I] time="2024-02-23T19:08:08.061116255Z" level=debug msg="Calling GET /v1.44/events?filters=%7B%22type%22%3A%7B%22container%22%3Atrue%7D%7D"
[2024-02-23T19:08:08.061887171Z][dockerd][I] time="2024-02-23T19:08:08.061831213Z" level=debug msg="Calling GET /v1.44/containers/json?all=1"
[2024-02-23T19:08:08.062861588Z][dockerd][I] time="2024-02-23T19:08:08.061832921Z" level=debug msg="Calling GET /v1.44/containers/json?all=1"
[2024-02-23T19:08:08.070751296Z][dockerd][I] time="2024-02-23T19:08:08.070596380Z" level=debug msg="Calling GET /v1.44/containers/6f4c4e728271/stats?stream=0"
[2024-02-23T19:08:08.070755796Z][dockerd][I] time="2024-02-23T19:08:08.070645005Z" level=debug msg="Calling GET /v1.44/containers/a831d87768d9/stats?stream=0"
[2024-02-23T19:08:08.070756130Z][dockerd][I] time="2024-02-23T19:08:08.070666338Z" level=debug msg="Calling GET /v1.44/containers/6f4c4e728271/stats?stream=0"
[2024-02-23T19:08:08.070757338Z][dockerd][I] time="2024-02-23T19:08:08.070678338Z" level=debug msg="Calling GET /v1.44/containers/082cc58691be/stats?stream=0"
[2024-02-23T19:08:08.070757588Z][dockerd][I] time="2024-02-23T19:08:08.070600671Z" level=debug msg="Calling GET /v1.44/containers/f066f053e584/stats?stream=0"
[...] one per container
[2024-02-23T19:08:09.589168172Z][dockerd][I] time="2024-02-23T19:08:09.588888922Z" level=debug msg="Client context cancelled, stop sending events"
[2024-02-23T19:08:09.590033380Z][dockerd][I] time="2024-02-23T19:08:09.588950505Z" level=debug msg="Client context cancelled, stop sending events"
It's clear something is calling into the daemon and it's appx every 10s, anyone got any advice as to how to trace what? This happens pretty continually - I don't have any IDE open at the time. What I do have is crowdstrike/falcon going on which would be suspect, but I've been told that there's nothing set up to monitor docker. I won't believe it without proof of what is actually submitting these commands though.
It might be good to know if everyone else having these problems also sees these kind of logs every 10 or so seconds. I cannot think what needs to ping docker every 10 flipping seconds!
same here. how can i trace what calls the daemon?
[2024-03-19T06:07:02.003660213Z][dockerd][I] time="2024-03-19T06:07:02.003424338Z" level=debug msg="Calling GET /v1.44/images/json?filters=%7B%22reference%22%3A%7B%22eclipse-temurin%22%3Atrue%7D%7D" [2024-03-19T06:07:02.009916588Z][dockerd][I] time="2024-03-19T06:07:02.009833671Z" level=debug msg="Calling HEAD /_ping" [2024-03-19T06:07:02.011109838Z][dockerd][I] time="2024-03-19T06:07:02.011012546Z" level=debug msg="Calling GET /v1.44/images/json?filters=%7B%22reference%22%3A%7B%22tomcat%22%3Atrue%7D%7D" [2024-03-19T06:07:02.013603713Z][dockerd][I] time="2024-03-19T06:07:02.013543005Z" level=debug msg="Calling HEAD /_ping" [2024-03-19T06:07:02.014141546Z][dockerd][I] time="2024-03-19T06:07:02.014068088Z" level=debug msg="Calling GET /v1.44/images/json?filters=%7B%22reference%22%3A%7B%22gradle%22%3Atrue%7D%7D" [2024-03-19T06:07:02.015923463Z][dockerd][I] time="2024-03-19T06:07:02.015867338Z" level=debug msg="Calling HEAD /_ping" [2024-03-19T06:07:02.016779338Z][dockerd][I] time="2024-03-19T06:07:02.016731088Z" level=debug msg="Calling GET /v1.44/images/json?filters=%7B%22reference%22%3A%7B%22maven%22%3Atrue%7D%7D" [2024-03-19T06:07:02.018620463Z][dockerd][I] time="2024-03-19T06:07:02.018565046Z" level=debug msg="Calling HEAD /_ping" [2024-03-19T06:07:02.019354296Z][dockerd][I] time="2024-03-19T06:07:02.019309046Z" level=debug msg="Calling GET /v1.44/images/json?filters=%7B%22reference%22%3A%7B%22postgres%22%3Atrue%7D%7D" [2024-03-19T06:07:02.021619255Z][dockerd][I] time="2024-03-19T06:07:02.021542921Z" level=debug msg="Calling HEAD /_ping" [2024-03-19T06:07:02.022291880Z][dockerd][I] time="2024-03-19T06:07:02.022241546Z" level=debug msg="Calling GET /v1.44/images/json?filters=%7B%22reference%22%3A%7B%22nginx%22%3Atrue%7D%7D"
Yours looks a little different to mine @UriZafrir, you have filters like:
GET /v1.44/images/json?filters={"reference":{"eclipse-temurin":true}}
GET /v1.44/images/json?filters={"reference":{"tomcat":true}}
GET /v1.44/images/json?filters={"reference":{"gradle":true}}
GET /v1.44/images/json?filters={"reference":{"maven":true}}
GET /v1.44/images/json?filters={"reference":{"postgres":true}}
GET /v1.44/images/json?filters={"reference":{"nginx":true}}
So, something looking for very specific things - at a guess, you have an IDE plugin (are you using eclipse) or some other system that is looking for things it knows how to handle. What applications are running at the same time as these logs?
I've done a little bit of process watching. With the Activity Monitor update frequency increased to 1s, I can see a process com.docker.cli
appearing after pausing docker. The parent process? Docker Desktop
.
Docker desktop (the app) was not open on my screen. With it open, I can see "docker is paused" on the main screen, but then it springs to life when something else pings the api.
Given my call profile of a ping, listing all type=container, then calling stats for each one, I believe something inside docker desktop is not respecting it being paused, and polling for stats. I assumed the 'Resource Usage' extension (docker 1st party extension), but updating/removing that did not seem to have an effect.
But.
I've since restarted docker desktop without the extension, and it looks like that might have actually had an effect. I'm going to monitor over the next couple of days, but this may boil down to docker extensions needing to be forcefully stopped when docker desktop is paused, or at least be prevented from calling the cli. At least, it's an update to Resource Usage.
EDIT: I also seemed unable to adjust the resource usage poll frequency - it would let me select 5m in the dropdown, but never applied it - when I navigated away and back it had just reset to 5s.
So, anyone else with this problem (docker desktop being unapused every 5-10s) I would suggest trying:
If that works for you, give this a 👍 so we can see if that's widespread, or just me.
Hi Spiral, that might be a good guess. I though it solved the problem but I'm not sure because after 10 minutes it resumed again after opening vscode. I'm using VScode. I also have a kind cluster installed with crossplane on it using docker. I've closed VScode and it seems docker engine never resumes! After opening VScode docker resumed. I thought my kubernetes extension polls docker so I've disabled it and suddenly for 10 minutes the docker didn't resume. I will try to find more details. other extensions installed are python, yaml and git graph. it used to resume faster, so I guess there might be another thing querying docker.
I still don't understand what to do with these logs to understand what is causing this behavior. it seems the logs got shorter, this is the loop (dropped all non necessary text):
"Calling HEAD /_ping" "Calling HEAD /_ping" "Calling GET /v1.44/version" "Calling GET /v1.44/version" "Calling GET /v1.44/events?filters={"type":{"container":true}}" "Calling GET /v1.44/events?filters={"type":{"container":true}}" "Calling GET /v1.44/containers/json?all=1" "Calling GET /v1.44/containers/json?all=1" "Calling GET /v1.44/containers/00330a101eb6/stats?stream=0" "Calling GET /v1.44/containers/00330a101eb6/stats?stream=0"
Same issue
@UriZafrir your logs now look exactly like mine did! Can you try what I suggested and remove all docker desktop extensions and restart docker completely. I think there's a zombie process for at least resource usage that persists when the extension is removed, and the system needs a restart to be able to purge it.
@burfot same suggestion - quit all your IDE's and apps other than docker, see if it persists, then remove all your extensions, restart and try again.
I have no extensions installed. I have restarted the PC a couple times.
Kind regards, Paul
On 2024-03-25, at 23:54, David Bennington @.***> wrote:
@UriZafrirhttps://github.com/UriZafrir your logs now look exactly like mine did! Can you try what I suggested and remove all docker desktop extensions and restart docker completely. I think there's a zombie process for at least resource usage that persists when the extension is removed, and the system needs a restart to be able to purge it.
@burfothttps://github.com/burfot same suggestion - quit all your IDE's and apps other than docker, see if it persists, then remove all your extensions, restart and try again.
— Reply to this email directly, view it on GitHubhttps://github.com/docker/for-mac/issues/6908#issuecomment-2017723682, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ALRNCC4KGN6AXP7SPAQQUULYZ767DAVCNFSM6AAAAAA2EIKUAGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAMJXG4ZDGNRYGI. You are receiving this because you were mentioned.Message ID: @.***>
I also had no extensions. I gave up, went for rancher desktop or colima with docker.
Description
Several times now i've noticed that docker is running, despite having previously paused it. It appears that some time after pausing docker (approx 5 minutes or so?), it inexplicably resumes itself and is running again.
Reproduce
In the menu bar, click on the Docker Desktop icon, then click Pause. After some time, clicking back on this icon it appears that at some point docker is running again, so has resumed by itself.
Expected behavior
After pause, docker should remain paused until the user explicitly opts to resume.
docker version
docker info
Diagnostics ID
9A4D113C-F0C8-4CC0-8D39-70FD8C991DF4/20230710092805
Additional Info
No response